God's Eye
Initial Launch.
Eight weeks with the Broadcast team. We build the first real version around the sources and events that matter most, put it in your hands, and learn together what deserves to become permanent.
A new intelligence product, not another dashboard.
God's Eye is ingestion infrastructure, scraping, source normalization, narrative detection, ranking, source traceability and a new interface. It can be prototyped quickly — but the risk and the product value are both far higher than a single-workflow build. So the sprint is scoped around proving it on real sources, in a real event, with your team.
Built around what your team asked for.
In our session with Broadcast leadership, seven things came up again and again. Here is exactly how the sprint answers each one.
React while the event is happening
The event pilot runs during the live event window, not just after — so the team can act on a rising story mid-event, while it still changes the show.
Separate esports talk from game talk
Conversations are classified as competitive-narrative or general game talk — and the game-side findings are packaged so they can be passed to the game team.
One morning read
Wake up to what everyone is talking about, where it came from, and the top creator posts from the last twelve to sixteen hours.
Co-streamers at scale
Live chat paired with what the streamer is saying — “this streamer was excited about the new mode, here's the VOD timestamp” — so it can be clipped or put in the broadcast. Calibrated on one named co-streamer first.
Regional language views
Broadcast runs in over a dozen languages. Regional teams get the heat in their own region, in their own language.
Attribute spikes to your own posts
Conversation spikes are linked to Riot's own content drops, so “we posted a player feature that day” sits next to the topic it moved.
One system instead of scattered dashboards
God's Eye consolidates what internal dashboards and agency reports split across tools — and it's the only system that keeps adding more, customized to each team.
The same system used to cost millions.
Listening to an entire community and turning it into stories is not a new idea — it is an old, expensive one. What changed is the cost curve. The hard parts (reading everything, understanding context, explaining why it matters) used to be custom research projects. Now they are infrastructure.
Cost and time to a working narrative engine
Industry estimates · hover a yearA research team, hand-built crawlers, and models trained from scratch.
Cloud pipelines and sentiment APIs — still a dashboard, not an answer.
Models read, cluster and explain. The work is now judgement and interface.
Entertainment has done this before — at studio budgets.
Netflix built recommendation and audience-signal systems that shaped what got greenlit. Major studios published research on reading trailer reactions and audience response to predict interest. Sports leagues run their own listening rooms to know what fans are talking about before broadcast goes live. This is no longer experimental — the biggest names in entertainment have already proven the playbook works.
That is exactly why Riot needs this now. When audiences move faster than production cycles, the team that hears the conversation first owns the story. Esports is the last major fan ecosystem without a system like this pointed at it — and whoever builds it first sets the standard everyone else follows.
Eight weeks is now a reasonable ask.
The expensive layers — language understanding, clustering, summarisation — are bought, not built. What is left is the part that actually needs Riot: which sources matter, which stories deserve airtime, and what the Broadcast team should do with them. That is exactly what the sprint spends its time on.
Eight weeks. Eight things delivered.
Source layer
An initial set of high-value esports sources: Reddit, VLR, YouTube, selected creator and co-streamer sources with live chat, news and community sites, and other technically viable public sources.
Collection system
Scheduled scraping and ingestion, source normalization, deduplication, and basic health monitoring so the pipeline stays honest.
Story intelligence
Detect recurring themes, emerging narratives, sentiment and context shifts, and unusual spikes across sources — with spikes attributable to Riot's own content drops.
God's Eye view
A dashboard showing the most important community narratives, where they originated, evidence and source links, momentum, and why each story matters.
Event pilot
A real Riot event or competition as the proving ground — tuned during the live event window, not just after, alongside the Broadcast team.
Editorial controls
Riot can inspect sources, dismiss noise, flag useful stories, and help tune what the engine treats as important.
Iteration
Weekly working sessions with the team and rapid changes throughout the eight weeks.
Launch output
A working hosted product, documentation, and a recommendation for what should become Phase 2.
The engine is proven.
The interface is yours to shape.
What you've seen running is a working intelligence engine — not a finished product design. God's Eye gets designed to match the quality of the tools Broadcast already lives in, and the tone of every insight gets written the way your team wants to read it.
It can be tailored to exactly what you want. That requires knowing what you want — or spending the time with us to figure it out.
Live ingestion, narrative detection and traceable evidence already running on real Valorant data.
The current view is a working harness. The product design comes in the sprint.
We design to match the apps Broadcast uses today — not a generic dashboard.
Insight tone, length and framing tuned to how your team reads and acts.
Collect. Detect. Show. Prove.
Sources & ingestion
Lock the source list with Broadcast, stand up scheduled collection, normalization, deduplication and health monitoring.
Narrative detection
Theme clustering, sentiment and context shifts, spike detection, and the ranking logic that decides what matters.
The God's Eye view
The dashboard, evidence and source links, momentum and confidence, plus the editorial controls for dismissing noise.
Event pilot & launch
Run it live against a real Riot event, tune with the team, hand over the hosted product, documentation and the Phase 2 recommendation.
We are not promising the internet.
- A working first version around the highest-value sources
- A real Riot use case, piloted on a real event
- Narrative detection with full source traceability
- Production-grade Twitch chat capture — the full live-chat firehose, live during events
- Global language coverage — regional teams get the heat in their own language
- An interface and insight tone designed with Broadcast — to the quality of the tools you use today
- Editorial controls so Broadcast can tune the engine
- Hosted product, documentation, and a Phase 2 recommendation
- Unlimited sources or every social network
- An enterprise social-listening replacement
- Historical internet archives
- Sources that actively prohibit scraping
Those are exactly how a fixed-price engagement accidentally turns into a $200K platform.
Easy to buy. Hard to fake.
The conversation is simply: give us eight weeks with the Broadcast team, and we'll put a working first version in your hands.
Eight weeks with the Broadcast team, a working God's Eye in your hands, and a clear view of what deserves to become permanent — a rare chance to shape the system before it scales.
You are buying the first two months, not setting the lifetime value of God's Eye.
The scope is deliberately bounded so this does not quietly become a $200K platform.
Adding features, improving them, or iterating is a monthly working agreement after the launch — not part of this fixed price.
Automations, or connecting God's Eye to Riot's existing systems, is a separate scope of work.
You decide what becomes permanent.
Iterate together: work with us monthly to add features, improve them, or keep tuning what the engine surfaces.
Automate & integrate: automations, or connecting God's Eye to existing Riot systems — scoped and quoted separately.
Or stop — you keep the working product, the documentation and the research.
Every next step is a separate decision, with something real in front of you.
Maintenance scales with the data — not with how many creators you watch.
Monitor 12 creators or 125 — the price doesn't care about source count. What it scales with is the volume of conversations the engine reads and analyzes each month. One monthly price covers everything — hosting, data collection, AI analysis, monitoring and fixes. No separate bills, no surprise invoices, nothing passed through.
Anything one person says that the engine reads. A comment under an FNS video. A Reddit post. A reply inside that post's thread. A single live-chat message during a co-stream. The spoken words of a video (its transcript). One FNS video with 400 comments counts as one video plus 400 conversations — that's it, no tricks.
Six platforms are listening today — Reddit, YouTube, Twitch, Kick, VLR and X — and adding more doesn't change the logic, only the volume that flows through.
Start at 400K conversations/month ($4,300/mo) — enough headroom for Valorant plus your creator and co-stream coverage in normal weeks and event weeks alike. You can move up or down a band any month as you learn what the team actually reads.
What the entire listening layer costs per day at this volume — six platforms, every language, around the clock, for less than a single production line item.
Of the conversation read, not sampled. A human team could skim a fraction of 13,333 daily messages — the engine reads every one of them, in every language, before your morning stand-up.
In surprise bills. Data collection, AI analysis, hosting, scraper repairs when platforms change — it's all inside the one number. Cancel any month and everything built stays with Riot.
For comparison: reading 13,333 messages a day by hand would take roughly 9 full-time readers across time zones and languages — and they'd still miss the overnight co-stream chatter. The engine covers all of it before your morning stand-up, at a fraction of that staffing cost, whatever volume band you choose.
Servers, database, storage and bandwidth to keep God's Eye live and fast.
The model usage that reads every conversation, detects narratives and writes the briefs.
Source monitoring, scraper repairs when platforms change, bug fixes and quiet upkeep.
A weekly touchpoint with the team — what's working, what's noisy, what to tune.
Unlimited sources and creators inside your volume band — one price, everything included. New features, automations, or connections to Riot's existing systems are quoted separately — maintenance keeps God's Eye running; it doesn't grow it. Cancel any month: the product, documentation and research stay with Riot.
Start the God's Eye Launch Sprint.
Eight weeks, $49,500, one real event as the proving ground. At the end you have a working hosted product and a clear recommendation for what comes next.